Breathing motion correction in single photon emission computed tomography

Single Photon Emission Computed Tomography (SPECT) is a medical imaging modality that allows to obtain the 3D distribution of a gamma-emitting tracer injected into the patient. It is particularly used in oncology for the diagnosis and localization of tumors. More recently, with the development of selective internal radiotherapy (SIRT), it has also become an important tool for the targeted treatment of cancer cells. In this case, the accuracy of the SPECT images influences the treatment received by the patient. As with other imaging modalities, respiratory motion creates blur and artifacts in the reconstructed images. The problem of respiratory motion has been studied in Positron Emission Tomography (PET) and cardiac SPECT but barely for other SPECT applications. The objective of this thesis is to develop methods to take into account respiratory motion for clinical SPECT systems and to study their influence on liver radioembolization. The first method proposed is a respiration-gated reconstruction. In this approach, a respiratory signal is automatically extracted from the measured projections and used to sort the measured projections in small temporal respiratory gated frames with minimal motion. Then, the sorted projections are reconstructed phase per phase yielding a series of motion-free volumetric images also called 4D SPECT image. However, the lower photon count in the projection data of each individual time frame compared to conventional reconstruction leads to SPECT images with a poorer signal-to-noise ratio. A second approach to compensate for respiratory motion while using all the available data was also developed. For this, the projections are sorted according to the respiratory signal as before but the motion is compensated before reconstruction. A 2D affine motion was estimated between projections at different phases. The transformation parameters were used to re-bin the list-mode data into one set of compensated projections that was then used to reconstruct a 3D motion-compensated SPECT image using all available events of the list-mode data. Each of these approaches was evaluated on simulated and real data. The real data came from SPECT acquisitions of a dynamic phantom but also from pre-treatment imaging of patients treated with liver radioembolization. Our methods showed an increase in the activity measured in the tumors compared to conventional 3D reconstructions. For radioembolization, the results suggest a possible benefit on treatment planning. This work also highlights the importance of the choice of the attenuation map for attenuation correction and segmentation of motion corrected images. Among the perspectives opened by this work we find the in-depth study of the attenuation correction for motion corrected images but also the application of our methods on other SPECT applications. (author)